Predicting the risk of cancer recurrence or death for patients diagnosed in the middle stages of cancer remains a difficult problem. Survival outcomes are particularly variable for patients with Stage II (Duke's B) or Stage III (Duke's C) colorectal cancer, who represent the majority (855%) of the 150,000 colorectal cancer cases diagnosed each year in the US. Genetic and biochemical heterogeneity among these cancers is complex-not readily measured by any single marker. However, accurate prediction of the risk of cancer recurrence or death are critical to optimal treatment, particularly as many new types of adjuvant therapy for colorectal cancer reach clinical trials. The PI proposes to address this problem with an innovative approach derived from her research on proteinase expression patterns in colorectal cancers. Proteolytic enzymes activities contribute functionally to cancer invasion and metastasis and can be sensitive predictors of cancer behavior. The characterization of proteolytic activities in primary human colorectal cancers and patient-matched control mucosa have identified heterogeneity. Two markers, cathepsin B (CB) and matrix metalloproteinase (MMP-9) have been found that demonstrate particularly distinctive patterns of expression in colorectal cancer. The PI's preliminary studies indicate that the combined expression patterns for these two markers ("CB/MMP-9 profiles") may generate significant new prognostic information on risk of recurrence and death. Thus, the major research goal of this grant will be to test dual proteinase markers, CB and MMP-9, for their statistical value individually and interactively as predictors of colorectal cancer survival outcomes. They will fit a set of Cox proportional hazard models to test the extent to which the expression of either marker alone and, most importantly, of the combined CB/MMP-9 profiles, predicts the risk of cancer recurrence or death among 400 colorectal cancer patients at all stages. They will also specifically test the value of this dual marker system for patients with intermediate levels of tumor invasion. These studies are designed to evaluate new predictors of cancer outcome that may ultimately be utilized for the assignment of patients to optimal treatment protocols, with the long term goal being increased survival of colorectal patients.